I think this was due not to the SNAPSHOT repos but to the fact that we do not use
a specific revision for all the plugins involved.
If we disable the SNAPSHOT repos there will be other negative effects. For example
when we are on a SNAPSHOT like say 1.5.1-SNAPSHOT then the snapshot jars are
deployed to the apache snapshot repo. If a user checks out the apacheds trunk and
tries to build that instead of building the svn area with the dependent projects then
those snapshots from shared and daemon will not be resolved. Even worse if a user
checks out apacheds/trunk/server-installers and tries to build that then it will fail if
the SNAPSHOT repos are not visible.
I think we just have to make sure we assign a specific revision to all the plugins we
use instead of not including a revision for a plugin. This will prevent the use of
SNAPSHOT plugins I hope.
There is also a pluginManagement section in the pom that we need to populate so
we lock all plugins to a specific revision instead of allowing the use of SNAPSHOTs.
Do you guys think that these problems are from having the SNAPSHOT repos (regular maven repo and plugin repo) in the maven repo stack? I am basically in the habit of disabling the SNAPSHOT repos in any project I build because it does stuff like this. And yes I always disable the SNAPSHOT repo in ApacheDS pom as well :-)
Its always been in the back of my mind to ask why it is enabled, but I figured that it allowed someone to build part of the SVN project without building the entire thing or something along those lines.
ChrisOn 5/21/07, Alex Karasulu < email@example.com> wrote:Ahhh the unpredictability of Maven strikes again!
AlexOn 5/21/07, Stefan Zoerner < firstname.lastname@example.org> wrote:Emmanuel Lecharny wrote:
> I have tested it again after having removed the repo, and still working ...
I have removed the repo from my machine, rebuilt it all and you are
right: I just came back from lunch and now I have fresh installers.
Thanks for your help,